{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.2,
  "eval_steps": 2000,
  "global_step": 4000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "grad_norm": 6.03125,
      "learning_rate": 3.0000000000000004e-07,
      "loss": 0.5688,
      "step": 10
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.5,
      "learning_rate": 6.000000000000001e-07,
      "loss": 0.4803,
      "step": 20
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.6875,
      "learning_rate": 9e-07,
      "loss": 0.6168,
      "step": 30
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.40625,
      "learning_rate": 1.2000000000000002e-06,
      "loss": 0.5197,
      "step": 40
    },
    {
      "epoch": 0.0,
      "grad_norm": 7.0,
      "learning_rate": 1.5e-06,
      "loss": 0.5441,
      "step": 50
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.6875,
      "learning_rate": 1.8e-06,
      "loss": 0.6031,
      "step": 60
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.4375,
      "learning_rate": 2.1000000000000002e-06,
      "loss": 0.4926,
      "step": 70
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.4375,
      "learning_rate": 2.4000000000000003e-06,
      "loss": 0.5795,
      "step": 80
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.75,
      "learning_rate": 2.7e-06,
      "loss": 0.5599,
      "step": 90
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.125,
      "learning_rate": 3e-06,
      "loss": 0.5279,
      "step": 100
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.96875,
      "learning_rate": 3.3e-06,
      "loss": 0.5455,
      "step": 110
    },
    {
      "epoch": 0.01,
      "grad_norm": 7.09375,
      "learning_rate": 3.6e-06,
      "loss": 0.5655,
      "step": 120
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.25,
      "learning_rate": 3.9e-06,
      "loss": 0.4635,
      "step": 130
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.8125,
      "learning_rate": 4.2000000000000004e-06,
      "loss": 0.64,
      "step": 140
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.9375,
      "learning_rate": 4.5e-06,
      "loss": 0.4565,
      "step": 150
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.09375,
      "learning_rate": 4.800000000000001e-06,
      "loss": 0.5357,
      "step": 160
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.59375,
      "learning_rate": 5.1e-06,
      "loss": 0.4885,
      "step": 170
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.9375,
      "learning_rate": 5.4e-06,
      "loss": 0.5125,
      "step": 180
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.21875,
      "learning_rate": 5.7000000000000005e-06,
      "loss": 0.4882,
      "step": 190
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.625,
      "learning_rate": 6e-06,
      "loss": 0.4504,
      "step": 200
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.15625,
      "learning_rate": 6.3e-06,
      "loss": 0.404,
      "step": 210
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.1875,
      "learning_rate": 6.6e-06,
      "loss": 0.4566,
      "step": 220
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.125,
      "learning_rate": 6.900000000000001e-06,
      "loss": 0.4597,
      "step": 230
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.0625,
      "learning_rate": 7.2e-06,
      "loss": 0.3678,
      "step": 240
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.09375,
      "learning_rate": 7.5e-06,
      "loss": 0.5107,
      "step": 250
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.15625,
      "learning_rate": 7.8e-06,
      "loss": 0.524,
      "step": 260
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.78125,
      "learning_rate": 8.1e-06,
      "loss": 0.4228,
      "step": 270
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.71875,
      "learning_rate": 8.400000000000001e-06,
      "loss": 0.4605,
      "step": 280
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.4375,
      "learning_rate": 8.7e-06,
      "loss": 0.4211,
      "step": 290
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.84375,
      "learning_rate": 9e-06,
      "loss": 0.438,
      "step": 300
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.375,
      "learning_rate": 9.3e-06,
      "loss": 0.4466,
      "step": 310
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.265625,
      "learning_rate": 9.600000000000001e-06,
      "loss": 0.3544,
      "step": 320
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.703125,
      "learning_rate": 9.9e-06,
      "loss": 0.3532,
      "step": 330
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.25,
      "learning_rate": 1.02e-05,
      "loss": 0.355,
      "step": 340
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.25,
      "learning_rate": 1.05e-05,
      "loss": 0.3628,
      "step": 350
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.9375,
      "learning_rate": 1.08e-05,
      "loss": 0.3914,
      "step": 360
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.3125,
      "learning_rate": 1.11e-05,
      "loss": 0.4033,
      "step": 370
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.625,
      "learning_rate": 1.1400000000000001e-05,
      "loss": 0.329,
      "step": 380
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.75,
      "learning_rate": 1.1700000000000001e-05,
      "loss": 0.3501,
      "step": 390
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.21875,
      "learning_rate": 1.2e-05,
      "loss": 0.3402,
      "step": 400
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.703125,
      "learning_rate": 1.2299999999999999e-05,
      "loss": 0.3807,
      "step": 410
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.4375,
      "learning_rate": 1.26e-05,
      "loss": 0.38,
      "step": 420
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.1875,
      "learning_rate": 1.29e-05,
      "loss": 0.3419,
      "step": 430
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.96875,
      "learning_rate": 1.32e-05,
      "loss": 0.4231,
      "step": 440
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.4375,
      "learning_rate": 1.3500000000000001e-05,
      "loss": 0.3869,
      "step": 450
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.375,
      "learning_rate": 1.3800000000000002e-05,
      "loss": 0.3628,
      "step": 460
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.59375,
      "learning_rate": 1.4099999999999999e-05,
      "loss": 0.3268,
      "step": 470
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.15625,
      "learning_rate": 1.44e-05,
      "loss": 0.4242,
      "step": 480
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.40625,
      "learning_rate": 1.47e-05,
      "loss": 0.3388,
      "step": 490
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.46875,
      "learning_rate": 1.5e-05,
      "loss": 0.3188,
      "step": 500
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.03125,
      "learning_rate": 1.4984210526315789e-05,
      "loss": 0.3138,
      "step": 510
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.9375,
      "learning_rate": 1.496842105263158e-05,
      "loss": 0.3381,
      "step": 520
    },
    {
      "epoch": 0.03,
      "grad_norm": 7.0625,
      "learning_rate": 1.4952631578947368e-05,
      "loss": 0.3579,
      "step": 530
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.40625,
      "learning_rate": 1.4936842105263158e-05,
      "loss": 0.308,
      "step": 540
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.125,
      "learning_rate": 1.4921052631578947e-05,
      "loss": 0.4211,
      "step": 550
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.375,
      "learning_rate": 1.4905263157894737e-05,
      "loss": 0.3596,
      "step": 560
    },
    {
      "epoch": 0.03,
      "grad_norm": 7.4375,
      "learning_rate": 1.4889473684210526e-05,
      "loss": 0.3154,
      "step": 570
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.28125,
      "learning_rate": 1.4873684210526315e-05,
      "loss": 0.3132,
      "step": 580
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.9375,
      "learning_rate": 1.4857894736842107e-05,
      "loss": 0.3773,
      "step": 590
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.484375,
      "learning_rate": 1.4842105263157895e-05,
      "loss": 0.3634,
      "step": 600
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.875,
      "learning_rate": 1.4826315789473686e-05,
      "loss": 0.2933,
      "step": 610
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.515625,
      "learning_rate": 1.4810526315789474e-05,
      "loss": 0.3159,
      "step": 620
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.3125,
      "learning_rate": 1.4794736842105265e-05,
      "loss": 0.3357,
      "step": 630
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.875,
      "learning_rate": 1.4778947368421053e-05,
      "loss": 0.3478,
      "step": 640
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.0625,
      "learning_rate": 1.4763157894736842e-05,
      "loss": 0.3873,
      "step": 650
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.03125,
      "learning_rate": 1.4747368421052632e-05,
      "loss": 0.2899,
      "step": 660
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.78125,
      "learning_rate": 1.4731578947368421e-05,
      "loss": 0.3353,
      "step": 670
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.0625,
      "learning_rate": 1.4715789473684211e-05,
      "loss": 0.3488,
      "step": 680
    },
    {
      "epoch": 0.03,
      "grad_norm": 8.625,
      "learning_rate": 1.47e-05,
      "loss": 0.3456,
      "step": 690
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.125,
      "learning_rate": 1.468421052631579e-05,
      "loss": 0.3656,
      "step": 700
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.0625,
      "learning_rate": 1.4668421052631579e-05,
      "loss": 0.3494,
      "step": 710
    },
    {
      "epoch": 0.04,
      "grad_norm": 5.28125,
      "learning_rate": 1.4652631578947367e-05,
      "loss": 0.2928,
      "step": 720
    },
    {
      "epoch": 0.04,
      "grad_norm": 7.625,
      "learning_rate": 1.4636842105263158e-05,
      "loss": 0.391,
      "step": 730
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.21875,
      "learning_rate": 1.4621052631578946e-05,
      "loss": 0.3042,
      "step": 740
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.625,
      "learning_rate": 1.4605263157894737e-05,
      "loss": 0.2822,
      "step": 750
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.0625,
      "learning_rate": 1.4589473684210527e-05,
      "loss": 0.3103,
      "step": 760
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.9375,
      "learning_rate": 1.4573684210526317e-05,
      "loss": 0.2982,
      "step": 770
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.28125,
      "learning_rate": 1.4557894736842106e-05,
      "loss": 0.2873,
      "step": 780
    },
    {
      "epoch": 0.04,
      "grad_norm": 11.375,
      "learning_rate": 1.4542105263157895e-05,
      "loss": 0.346,
      "step": 790
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.71875,
      "learning_rate": 1.4526315789473685e-05,
      "loss": 0.3055,
      "step": 800
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.96875,
      "learning_rate": 1.4510526315789474e-05,
      "loss": 0.2721,
      "step": 810
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.28125,
      "learning_rate": 1.4494736842105264e-05,
      "loss": 0.2343,
      "step": 820
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.90625,
      "learning_rate": 1.4478947368421053e-05,
      "loss": 0.2855,
      "step": 830
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.875,
      "learning_rate": 1.4463157894736843e-05,
      "loss": 0.2685,
      "step": 840
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.28125,
      "learning_rate": 1.4447368421052632e-05,
      "loss": 0.2675,
      "step": 850
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.84375,
      "learning_rate": 1.443157894736842e-05,
      "loss": 0.3448,
      "step": 860
    },
    {
      "epoch": 0.04,
      "grad_norm": 5.5625,
      "learning_rate": 1.441578947368421e-05,
      "loss": 0.2843,
      "step": 870
    },
    {
      "epoch": 0.04,
      "grad_norm": 5.4375,
      "learning_rate": 1.44e-05,
      "loss": 0.3069,
      "step": 880
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.75,
      "learning_rate": 1.438421052631579e-05,
      "loss": 0.2324,
      "step": 890
    },
    {
      "epoch": 0.04,
      "grad_norm": 5.8125,
      "learning_rate": 1.4368421052631578e-05,
      "loss": 0.3389,
      "step": 900
    },
    {
      "epoch": 0.05,
      "grad_norm": 5.96875,
      "learning_rate": 1.4352631578947369e-05,
      "loss": 0.3082,
      "step": 910
    },
    {
      "epoch": 0.05,
      "grad_norm": 4.0625,
      "learning_rate": 1.4336842105263159e-05,
      "loss": 0.2889,
      "step": 920
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.109375,
      "learning_rate": 1.4321052631578948e-05,
      "loss": 0.2842,
      "step": 930
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.96875,
      "learning_rate": 1.4305263157894738e-05,
      "loss": 0.2695,
      "step": 940
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.875,
      "learning_rate": 1.4289473684210527e-05,
      "loss": 0.2021,
      "step": 950
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.03125,
      "learning_rate": 1.4273684210526317e-05,
      "loss": 0.3382,
      "step": 960
    },
    {
      "epoch": 0.05,
      "grad_norm": 5.09375,
      "learning_rate": 1.4257894736842106e-05,
      "loss": 0.3721,
      "step": 970
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.921875,
      "learning_rate": 1.4242105263157896e-05,
      "loss": 0.2715,
      "step": 980
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.71875,
      "learning_rate": 1.4226315789473685e-05,
      "loss": 0.2261,
      "step": 990
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.40625,
      "learning_rate": 1.4210526315789473e-05,
      "loss": 0.2718,
      "step": 1000
    },
    {
      "epoch": 0.05,
      "grad_norm": 5.09375,
      "learning_rate": 1.4194736842105264e-05,
      "loss": 0.355,
      "step": 1010
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.0625,
      "learning_rate": 1.4178947368421052e-05,
      "loss": 0.3133,
      "step": 1020
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.734375,
      "learning_rate": 1.4163157894736843e-05,
      "loss": 0.3427,
      "step": 1030
    },
    {
      "epoch": 0.05,
      "grad_norm": 5.5625,
      "learning_rate": 1.4147368421052631e-05,
      "loss": 0.2373,
      "step": 1040
    },
    {
      "epoch": 0.05,
      "grad_norm": 4.53125,
      "learning_rate": 1.4131578947368422e-05,
      "loss": 0.3505,
      "step": 1050
    },
    {
      "epoch": 0.05,
      "grad_norm": 4.21875,
      "learning_rate": 1.411578947368421e-05,
      "loss": 0.3211,
      "step": 1060
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.984375,
      "learning_rate": 1.4099999999999999e-05,
      "loss": 0.3006,
      "step": 1070
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.765625,
      "learning_rate": 1.408421052631579e-05,
      "loss": 0.31,
      "step": 1080
    },
    {
      "epoch": 0.05,
      "grad_norm": 5.6875,
      "learning_rate": 1.406842105263158e-05,
      "loss": 0.2641,
      "step": 1090
    },
    {
      "epoch": 0.06,
      "grad_norm": 9.375,
      "learning_rate": 1.405263157894737e-05,
      "loss": 0.3475,
      "step": 1100
    },
    {
      "epoch": 0.06,
      "grad_norm": 6.09375,
      "learning_rate": 1.4036842105263158e-05,
      "loss": 0.2827,
      "step": 1110
    },
    {
      "epoch": 0.06,
      "grad_norm": 5.46875,
      "learning_rate": 1.4021052631578949e-05,
      "loss": 0.2408,
      "step": 1120
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.84375,
      "learning_rate": 1.4005263157894737e-05,
      "loss": 0.2428,
      "step": 1130
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.375,
      "learning_rate": 1.3989473684210526e-05,
      "loss": 0.2533,
      "step": 1140
    },
    {
      "epoch": 0.06,
      "grad_norm": 4.84375,
      "learning_rate": 1.3973684210526316e-05,
      "loss": 0.3098,
      "step": 1150
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.875,
      "learning_rate": 1.3957894736842105e-05,
      "loss": 0.3235,
      "step": 1160
    },
    {
      "epoch": 0.06,
      "grad_norm": 4.6875,
      "learning_rate": 1.3942105263157895e-05,
      "loss": 0.2959,
      "step": 1170
    },
    {
      "epoch": 0.06,
      "grad_norm": 4.125,
      "learning_rate": 1.3926315789473684e-05,
      "loss": 0.2873,
      "step": 1180
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.515625,
      "learning_rate": 1.3910526315789474e-05,
      "loss": 0.3158,
      "step": 1190
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.734375,
      "learning_rate": 1.3894736842105263e-05,
      "loss": 0.3004,
      "step": 1200
    },
    {
      "epoch": 0.06,
      "grad_norm": 4.34375,
      "learning_rate": 1.3878947368421052e-05,
      "loss": 0.3596,
      "step": 1210
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.453125,
      "learning_rate": 1.3863157894736842e-05,
      "loss": 0.2345,
      "step": 1220
    },
    {
      "epoch": 0.06,
      "grad_norm": 4.6875,
      "learning_rate": 1.384736842105263e-05,
      "loss": 0.3332,
      "step": 1230
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.921875,
      "learning_rate": 1.3831578947368421e-05,
      "loss": 0.2898,
      "step": 1240
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.890625,
      "learning_rate": 1.3815789473684211e-05,
      "loss": 0.2854,
      "step": 1250
    },
    {
      "epoch": 0.06,
      "grad_norm": 4.03125,
      "learning_rate": 1.3800000000000002e-05,
      "loss": 0.2932,
      "step": 1260
    },
    {
      "epoch": 0.06,
      "grad_norm": 4.03125,
      "learning_rate": 1.378421052631579e-05,
      "loss": 0.3153,
      "step": 1270
    },
    {
      "epoch": 0.06,
      "grad_norm": 6.21875,
      "learning_rate": 1.3768421052631579e-05,
      "loss": 0.2947,
      "step": 1280
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.234375,
      "learning_rate": 1.375263157894737e-05,
      "loss": 0.2211,
      "step": 1290
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.15625,
      "learning_rate": 1.3736842105263158e-05,
      "loss": 0.296,
      "step": 1300
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.21875,
      "learning_rate": 1.3721052631578948e-05,
      "loss": 0.3276,
      "step": 1310
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.46875,
      "learning_rate": 1.3705263157894737e-05,
      "loss": 0.2716,
      "step": 1320
    },
    {
      "epoch": 0.07,
      "grad_norm": 4.53125,
      "learning_rate": 1.3689473684210527e-05,
      "loss": 0.2471,
      "step": 1330
    },
    {
      "epoch": 0.07,
      "grad_norm": 6.5,
      "learning_rate": 1.3673684210526316e-05,
      "loss": 0.3197,
      "step": 1340
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.953125,
      "learning_rate": 1.3657894736842106e-05,
      "loss": 0.275,
      "step": 1350
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.90625,
      "learning_rate": 1.3642105263157895e-05,
      "loss": 0.2567,
      "step": 1360
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.890625,
      "learning_rate": 1.3626315789473684e-05,
      "loss": 0.2856,
      "step": 1370
    },
    {
      "epoch": 0.07,
      "grad_norm": 8.5,
      "learning_rate": 1.3610526315789474e-05,
      "loss": 0.3118,
      "step": 1380
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.8125,
      "learning_rate": 1.3594736842105263e-05,
      "loss": 0.2669,
      "step": 1390
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.375,
      "learning_rate": 1.3578947368421053e-05,
      "loss": 0.2718,
      "step": 1400
    },
    {
      "epoch": 0.07,
      "grad_norm": 6.90625,
      "learning_rate": 1.3563157894736842e-05,
      "loss": 0.3,
      "step": 1410
    },
    {
      "epoch": 0.07,
      "grad_norm": 7.59375,
      "learning_rate": 1.3547368421052634e-05,
      "loss": 0.2698,
      "step": 1420
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.578125,
      "learning_rate": 1.3531578947368422e-05,
      "loss": 0.2467,
      "step": 1430
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.09375,
      "learning_rate": 1.3515789473684211e-05,
      "loss": 0.2349,
      "step": 1440
    },
    {
      "epoch": 0.07,
      "grad_norm": 7.15625,
      "learning_rate": 1.3500000000000001e-05,
      "loss": 0.2565,
      "step": 1450
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.375,
      "learning_rate": 1.348421052631579e-05,
      "loss": 0.2881,
      "step": 1460
    },
    {
      "epoch": 0.07,
      "grad_norm": 4.9375,
      "learning_rate": 1.346842105263158e-05,
      "loss": 0.2777,
      "step": 1470
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.828125,
      "learning_rate": 1.3452631578947369e-05,
      "loss": 0.3411,
      "step": 1480
    },
    {
      "epoch": 0.07,
      "grad_norm": 4.28125,
      "learning_rate": 1.343684210526316e-05,
      "loss": 0.3102,
      "step": 1490
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.859375,
      "learning_rate": 1.3421052631578948e-05,
      "loss": 0.2269,
      "step": 1500
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.71875,
      "learning_rate": 1.3405263157894736e-05,
      "loss": 0.2404,
      "step": 1510
    },
    {
      "epoch": 0.08,
      "grad_norm": 5.21875,
      "learning_rate": 1.3389473684210527e-05,
      "loss": 0.2708,
      "step": 1520
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.171875,
      "learning_rate": 1.3373684210526315e-05,
      "loss": 0.2771,
      "step": 1530
    },
    {
      "epoch": 0.08,
      "grad_norm": 5.34375,
      "learning_rate": 1.3357894736842106e-05,
      "loss": 0.3431,
      "step": 1540
    },
    {
      "epoch": 0.08,
      "grad_norm": 4.21875,
      "learning_rate": 1.3342105263157894e-05,
      "loss": 0.2593,
      "step": 1550
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.375,
      "learning_rate": 1.3326315789473685e-05,
      "loss": 0.3764,
      "step": 1560
    },
    {
      "epoch": 0.08,
      "grad_norm": 4.5625,
      "learning_rate": 1.3310526315789473e-05,
      "loss": 0.2739,
      "step": 1570
    },
    {
      "epoch": 0.08,
      "grad_norm": 4.46875,
      "learning_rate": 1.3294736842105262e-05,
      "loss": 0.3306,
      "step": 1580
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.546875,
      "learning_rate": 1.3278947368421054e-05,
      "loss": 0.2687,
      "step": 1590
    },
    {
      "epoch": 0.08,
      "grad_norm": 5.5,
      "learning_rate": 1.3263157894736843e-05,
      "loss": 0.288,
      "step": 1600
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.390625,
      "learning_rate": 1.3247368421052633e-05,
      "loss": 0.2626,
      "step": 1610
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.0,
      "learning_rate": 1.3231578947368422e-05,
      "loss": 0.2635,
      "step": 1620
    },
    {
      "epoch": 0.08,
      "grad_norm": 4.1875,
      "learning_rate": 1.3215789473684212e-05,
      "loss": 0.2567,
      "step": 1630
    },
    {
      "epoch": 0.08,
      "grad_norm": 4.3125,
      "learning_rate": 1.32e-05,
      "loss": 0.2756,
      "step": 1640
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.21875,
      "learning_rate": 1.318421052631579e-05,
      "loss": 0.32,
      "step": 1650
    },
    {
      "epoch": 0.08,
      "grad_norm": 4.65625,
      "learning_rate": 1.316842105263158e-05,
      "loss": 0.2499,
      "step": 1660
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.125,
      "learning_rate": 1.3152631578947368e-05,
      "loss": 0.2084,
      "step": 1670
    },
    {
      "epoch": 0.08,
      "grad_norm": 5.0625,
      "learning_rate": 1.3136842105263159e-05,
      "loss": 0.2263,
      "step": 1680
    },
    {
      "epoch": 0.08,
      "grad_norm": 4.71875,
      "learning_rate": 1.3121052631578947e-05,
      "loss": 0.3104,
      "step": 1690
    },
    {
      "epoch": 0.09,
      "grad_norm": 7.21875,
      "learning_rate": 1.3105263157894738e-05,
      "loss": 0.3948,
      "step": 1700
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.921875,
      "learning_rate": 1.3089473684210526e-05,
      "loss": 0.2561,
      "step": 1710
    },
    {
      "epoch": 0.09,
      "grad_norm": 4.5,
      "learning_rate": 1.3073684210526315e-05,
      "loss": 0.2774,
      "step": 1720
    },
    {
      "epoch": 0.09,
      "grad_norm": 3.328125,
      "learning_rate": 1.3057894736842105e-05,
      "loss": 0.2165,
      "step": 1730
    },
    {
      "epoch": 0.09,
      "grad_norm": 3.046875,
      "learning_rate": 1.3042105263157894e-05,
      "loss": 0.2819,
      "step": 1740
    },
    {
      "epoch": 0.09,
      "grad_norm": 4.28125,
      "learning_rate": 1.3026315789473684e-05,
      "loss": 0.2111,
      "step": 1750
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.21875,
      "learning_rate": 1.3010526315789475e-05,
      "loss": 0.2293,
      "step": 1760
    },
    {
      "epoch": 0.09,
      "grad_norm": 3.53125,
      "learning_rate": 1.2994736842105265e-05,
      "loss": 0.2363,
      "step": 1770
    },
    {
      "epoch": 0.09,
      "grad_norm": 3.296875,
      "learning_rate": 1.2978947368421054e-05,
      "loss": 0.3236,
      "step": 1780
    },
    {
      "epoch": 0.09,
      "grad_norm": 4.53125,
      "learning_rate": 1.2963157894736842e-05,
      "loss": 0.2785,
      "step": 1790
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.265625,
      "learning_rate": 1.2947368421052633e-05,
      "loss": 0.238,
      "step": 1800
    },
    {
      "epoch": 0.09,
      "grad_norm": 3.828125,
      "learning_rate": 1.2931578947368421e-05,
      "loss": 0.2736,
      "step": 1810
    },
    {
      "epoch": 0.09,
      "grad_norm": 4.65625,
      "learning_rate": 1.2915789473684212e-05,
      "loss": 0.3267,
      "step": 1820
    },
    {
      "epoch": 0.09,
      "grad_norm": 4.03125,
      "learning_rate": 1.29e-05,
      "loss": 0.2644,
      "step": 1830
    },
    {
      "epoch": 0.09,
      "grad_norm": 4.875,
      "learning_rate": 1.288421052631579e-05,
      "loss": 0.2847,
      "step": 1840
    },
    {
      "epoch": 0.09,
      "grad_norm": 4.5625,
      "learning_rate": 1.2868421052631579e-05,
      "loss": 0.3246,
      "step": 1850
    },
    {
      "epoch": 0.09,
      "grad_norm": 3.328125,
      "learning_rate": 1.2852631578947368e-05,
      "loss": 0.2835,
      "step": 1860
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.9375,
      "learning_rate": 1.2836842105263158e-05,
      "loss": 0.2773,
      "step": 1870
    },
    {
      "epoch": 0.09,
      "grad_norm": 4.46875,
      "learning_rate": 1.2821052631578947e-05,
      "loss": 0.2967,
      "step": 1880
    },
    {
      "epoch": 0.09,
      "grad_norm": 3.5625,
      "learning_rate": 1.2805263157894737e-05,
      "loss": 0.3291,
      "step": 1890
    },
    {
      "epoch": 0.1,
      "grad_norm": 3.875,
      "learning_rate": 1.2789473684210526e-05,
      "loss": 0.2619,
      "step": 1900
    },
    {
      "epoch": 0.1,
      "grad_norm": 6.15625,
      "learning_rate": 1.2773684210526316e-05,
      "loss": 0.2467,
      "step": 1910
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.25,
      "learning_rate": 1.2757894736842106e-05,
      "loss": 0.224,
      "step": 1920
    },
    {
      "epoch": 0.1,
      "grad_norm": 7.09375,
      "learning_rate": 1.2742105263157895e-05,
      "loss": 0.2697,
      "step": 1930
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.765625,
      "learning_rate": 1.2726315789473685e-05,
      "loss": 0.245,
      "step": 1940
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.65625,
      "learning_rate": 1.2710526315789474e-05,
      "loss": 0.22,
      "step": 1950
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.828125,
      "learning_rate": 1.2694736842105264e-05,
      "loss": 0.3149,
      "step": 1960
    },
    {
      "epoch": 0.1,
      "grad_norm": 3.15625,
      "learning_rate": 1.2678947368421053e-05,
      "loss": 0.2345,
      "step": 1970
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.0,
      "learning_rate": 1.2663157894736843e-05,
      "loss": 0.2053,
      "step": 1980
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.84375,
      "learning_rate": 1.2647368421052632e-05,
      "loss": 0.2364,
      "step": 1990
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.0,
      "learning_rate": 1.263157894736842e-05,
      "loss": 0.3397,
      "step": 2000
    },
    {
      "epoch": 0.1,
      "eval_loss": 0.28045549988746643,
      "eval_runtime": 64.869,
      "eval_samples_per_second": 15.416,
      "eval_steps_per_second": 15.416,
      "step": 2000
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.9375,
      "learning_rate": 1.2615789473684211e-05,
      "loss": 0.2323,
      "step": 2010
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.625,
      "learning_rate": 1.26e-05,
      "loss": 0.2569,
      "step": 2020
    },
    {
      "epoch": 0.1,
      "grad_norm": 3.28125,
      "learning_rate": 1.258421052631579e-05,
      "loss": 0.2124,
      "step": 2030
    },
    {
      "epoch": 0.1,
      "grad_norm": 3.296875,
      "learning_rate": 1.2568421052631579e-05,
      "loss": 0.2332,
      "step": 2040
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.78125,
      "learning_rate": 1.2552631578947369e-05,
      "loss": 0.3277,
      "step": 2050
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.34375,
      "learning_rate": 1.2536842105263158e-05,
      "loss": 0.2017,
      "step": 2060
    },
    {
      "epoch": 0.1,
      "grad_norm": 3.59375,
      "learning_rate": 1.2521052631578946e-05,
      "loss": 0.2663,
      "step": 2070
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.53125,
      "learning_rate": 1.2505263157894737e-05,
      "loss": 0.2702,
      "step": 2080
    },
    {
      "epoch": 0.1,
      "grad_norm": 3.875,
      "learning_rate": 1.2489473684210527e-05,
      "loss": 0.2721,
      "step": 2090
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.1875,
      "learning_rate": 1.2473684210526317e-05,
      "loss": 0.2631,
      "step": 2100
    },
    {
      "epoch": 0.11,
      "grad_norm": 4.78125,
      "learning_rate": 1.2457894736842106e-05,
      "loss": 0.2694,
      "step": 2110
    },
    {
      "epoch": 0.11,
      "grad_norm": 3.609375,
      "learning_rate": 1.2442105263157896e-05,
      "loss": 0.2236,
      "step": 2120
    },
    {
      "epoch": 0.11,
      "grad_norm": 4.40625,
      "learning_rate": 1.2426315789473685e-05,
      "loss": 0.2442,
      "step": 2130
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.4375,
      "learning_rate": 1.2410526315789474e-05,
      "loss": 0.2605,
      "step": 2140
    },
    {
      "epoch": 0.11,
      "grad_norm": 4.09375,
      "learning_rate": 1.2394736842105264e-05,
      "loss": 0.2723,
      "step": 2150
    },
    {
      "epoch": 0.11,
      "grad_norm": 4.125,
      "learning_rate": 1.2378947368421053e-05,
      "loss": 0.2138,
      "step": 2160
    },
    {
      "epoch": 0.11,
      "grad_norm": 4.53125,
      "learning_rate": 1.2363157894736843e-05,
      "loss": 0.2413,
      "step": 2170
    },
    {
      "epoch": 0.11,
      "grad_norm": 6.28125,
      "learning_rate": 1.2347368421052631e-05,
      "loss": 0.2137,
      "step": 2180
    },
    {
      "epoch": 0.11,
      "grad_norm": 6.96875,
      "learning_rate": 1.2331578947368422e-05,
      "loss": 0.318,
      "step": 2190
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.40625,
      "learning_rate": 1.231578947368421e-05,
      "loss": 0.267,
      "step": 2200
    },
    {
      "epoch": 0.11,
      "grad_norm": 4.625,
      "learning_rate": 1.2299999999999999e-05,
      "loss": 0.2593,
      "step": 2210
    },
    {
      "epoch": 0.11,
      "grad_norm": 3.078125,
      "learning_rate": 1.228421052631579e-05,
      "loss": 0.2626,
      "step": 2220
    },
    {
      "epoch": 0.11,
      "grad_norm": 3.015625,
      "learning_rate": 1.2268421052631578e-05,
      "loss": 0.3095,
      "step": 2230
    },
    {
      "epoch": 0.11,
      "grad_norm": 7.03125,
      "learning_rate": 1.2252631578947368e-05,
      "loss": 0.2364,
      "step": 2240
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.84375,
      "learning_rate": 1.2236842105263159e-05,
      "loss": 0.3079,
      "step": 2250
    },
    {
      "epoch": 0.11,
      "grad_norm": 4.09375,
      "learning_rate": 1.2221052631578949e-05,
      "loss": 0.2233,
      "step": 2260
    },
    {
      "epoch": 0.11,
      "grad_norm": 3.796875,
      "learning_rate": 1.2205263157894738e-05,
      "loss": 0.3072,
      "step": 2270
    },
    {
      "epoch": 0.11,
      "grad_norm": 4.59375,
      "learning_rate": 1.2189473684210526e-05,
      "loss": 0.2579,
      "step": 2280
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.890625,
      "learning_rate": 1.2173684210526317e-05,
      "loss": 0.2265,
      "step": 2290
    },
    {
      "epoch": 0.12,
      "grad_norm": 3.875,
      "learning_rate": 1.2157894736842105e-05,
      "loss": 0.2932,
      "step": 2300
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.1875,
      "learning_rate": 1.2142105263157896e-05,
      "loss": 0.3118,
      "step": 2310
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.34375,
      "learning_rate": 1.2126315789473684e-05,
      "loss": 0.3196,
      "step": 2320
    },
    {
      "epoch": 0.12,
      "grad_norm": 3.59375,
      "learning_rate": 1.2110526315789475e-05,
      "loss": 0.3374,
      "step": 2330
    },
    {
      "epoch": 0.12,
      "grad_norm": 6.75,
      "learning_rate": 1.2094736842105263e-05,
      "loss": 0.2429,
      "step": 2340
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.5625,
      "learning_rate": 1.2078947368421052e-05,
      "loss": 0.2552,
      "step": 2350
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.4375,
      "learning_rate": 1.2063157894736842e-05,
      "loss": 0.25,
      "step": 2360
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.96875,
      "learning_rate": 1.2047368421052631e-05,
      "loss": 0.2549,
      "step": 2370
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.21875,
      "learning_rate": 1.2031578947368421e-05,
      "loss": 0.2079,
      "step": 2380
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.96875,
      "learning_rate": 1.201578947368421e-05,
      "loss": 0.269,
      "step": 2390
    },
    {
      "epoch": 0.12,
      "grad_norm": 7.28125,
      "learning_rate": 1.2e-05,
      "loss": 0.2658,
      "step": 2400
    },
    {
      "epoch": 0.12,
      "grad_norm": 6.0625,
      "learning_rate": 1.1984210526315789e-05,
      "loss": 0.3198,
      "step": 2410
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.0625,
      "learning_rate": 1.196842105263158e-05,
      "loss": 0.2094,
      "step": 2420
    },
    {
      "epoch": 0.12,
      "grad_norm": 6.84375,
      "learning_rate": 1.195263157894737e-05,
      "loss": 0.2877,
      "step": 2430
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.15625,
      "learning_rate": 1.1936842105263158e-05,
      "loss": 0.2458,
      "step": 2440
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.1875,
      "learning_rate": 1.1921052631578949e-05,
      "loss": 0.2186,
      "step": 2450
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.71875,
      "learning_rate": 1.1905263157894737e-05,
      "loss": 0.2657,
      "step": 2460
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.09375,
      "learning_rate": 1.1889473684210528e-05,
      "loss": 0.2838,
      "step": 2470
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.0,
      "learning_rate": 1.1873684210526316e-05,
      "loss": 0.283,
      "step": 2480
    },
    {
      "epoch": 0.12,
      "grad_norm": 3.34375,
      "learning_rate": 1.1857894736842105e-05,
      "loss": 0.2818,
      "step": 2490
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.0,
      "learning_rate": 1.1842105263157895e-05,
      "loss": 0.2842,
      "step": 2500
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.921875,
      "learning_rate": 1.1826315789473684e-05,
      "loss": 0.2352,
      "step": 2510
    },
    {
      "epoch": 0.13,
      "grad_norm": 8.25,
      "learning_rate": 1.1810526315789474e-05,
      "loss": 0.2655,
      "step": 2520
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.53125,
      "learning_rate": 1.1794736842105263e-05,
      "loss": 0.2457,
      "step": 2530
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.96875,
      "learning_rate": 1.1778947368421053e-05,
      "loss": 0.2637,
      "step": 2540
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.6875,
      "learning_rate": 1.1763157894736842e-05,
      "loss": 0.2808,
      "step": 2550
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.40625,
      "learning_rate": 1.174736842105263e-05,
      "loss": 0.2925,
      "step": 2560
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.6875,
      "learning_rate": 1.173157894736842e-05,
      "loss": 0.2698,
      "step": 2570
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.3125,
      "learning_rate": 1.171578947368421e-05,
      "loss": 0.2744,
      "step": 2580
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.546875,
      "learning_rate": 1.1700000000000001e-05,
      "loss": 0.2291,
      "step": 2590
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.15625,
      "learning_rate": 1.168421052631579e-05,
      "loss": 0.2584,
      "step": 2600
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.40625,
      "learning_rate": 1.166842105263158e-05,
      "loss": 0.197,
      "step": 2610
    },
    {
      "epoch": 0.13,
      "grad_norm": 7.96875,
      "learning_rate": 1.1652631578947369e-05,
      "loss": 0.2722,
      "step": 2620
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.15625,
      "learning_rate": 1.1636842105263158e-05,
      "loss": 0.2192,
      "step": 2630
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.375,
      "learning_rate": 1.1621052631578948e-05,
      "loss": 0.1978,
      "step": 2640
    },
    {
      "epoch": 0.13,
      "grad_norm": 6.78125,
      "learning_rate": 1.1605263157894737e-05,
      "loss": 0.2161,
      "step": 2650
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.3125,
      "learning_rate": 1.1589473684210527e-05,
      "loss": 0.2509,
      "step": 2660
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.546875,
      "learning_rate": 1.1573684210526316e-05,
      "loss": 0.2299,
      "step": 2670
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.09375,
      "learning_rate": 1.1557894736842106e-05,
      "loss": 0.286,
      "step": 2680
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.296875,
      "learning_rate": 1.1542105263157895e-05,
      "loss": 0.2221,
      "step": 2690
    },
    {
      "epoch": 0.14,
      "grad_norm": 4.6875,
      "learning_rate": 1.1526315789473683e-05,
      "loss": 0.2128,
      "step": 2700
    },
    {
      "epoch": 0.14,
      "grad_norm": 6.75,
      "learning_rate": 1.1510526315789474e-05,
      "loss": 0.2578,
      "step": 2710
    },
    {
      "epoch": 0.14,
      "grad_norm": 3.796875,
      "learning_rate": 1.1494736842105262e-05,
      "loss": 0.3137,
      "step": 2720
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8515625,
      "learning_rate": 1.1478947368421053e-05,
      "loss": 0.2952,
      "step": 2730
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.34375,
      "learning_rate": 1.1463157894736841e-05,
      "loss": 0.2424,
      "step": 2740
    },
    {
      "epoch": 0.14,
      "grad_norm": 7.03125,
      "learning_rate": 1.1447368421052632e-05,
      "loss": 0.1941,
      "step": 2750
    },
    {
      "epoch": 0.14,
      "grad_norm": 4.0625,
      "learning_rate": 1.1431578947368422e-05,
      "loss": 0.2441,
      "step": 2760
    },
    {
      "epoch": 0.14,
      "grad_norm": 4.71875,
      "learning_rate": 1.141578947368421e-05,
      "loss": 0.2584,
      "step": 2770
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.03125,
      "learning_rate": 1.1400000000000001e-05,
      "loss": 0.2769,
      "step": 2780
    },
    {
      "epoch": 0.14,
      "grad_norm": 6.875,
      "learning_rate": 1.138421052631579e-05,
      "loss": 0.2841,
      "step": 2790
    },
    {
      "epoch": 0.14,
      "grad_norm": 8.4375,
      "learning_rate": 1.136842105263158e-05,
      "loss": 0.2043,
      "step": 2800
    },
    {
      "epoch": 0.14,
      "grad_norm": 4.0625,
      "learning_rate": 1.1352631578947369e-05,
      "loss": 0.2642,
      "step": 2810
    },
    {
      "epoch": 0.14,
      "grad_norm": 4.90625,
      "learning_rate": 1.1336842105263159e-05,
      "loss": 0.1911,
      "step": 2820
    },
    {
      "epoch": 0.14,
      "grad_norm": 4.4375,
      "learning_rate": 1.1321052631578948e-05,
      "loss": 0.3403,
      "step": 2830
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.0,
      "learning_rate": 1.1305263157894736e-05,
      "loss": 0.224,
      "step": 2840
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.25,
      "learning_rate": 1.1289473684210527e-05,
      "loss": 0.2674,
      "step": 2850
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.1875,
      "learning_rate": 1.1273684210526315e-05,
      "loss": 0.2787,
      "step": 2860
    },
    {
      "epoch": 0.14,
      "grad_norm": 4.8125,
      "learning_rate": 1.1257894736842106e-05,
      "loss": 0.2718,
      "step": 2870
    },
    {
      "epoch": 0.14,
      "grad_norm": 6.09375,
      "learning_rate": 1.1242105263157894e-05,
      "loss": 0.2747,
      "step": 2880
    },
    {
      "epoch": 0.14,
      "grad_norm": 3.875,
      "learning_rate": 1.1226315789473685e-05,
      "loss": 0.2384,
      "step": 2890
    },
    {
      "epoch": 0.14,
      "grad_norm": 7.53125,
      "learning_rate": 1.1210526315789473e-05,
      "loss": 0.2541,
      "step": 2900
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.1875,
      "learning_rate": 1.1194736842105264e-05,
      "loss": 0.2794,
      "step": 2910
    },
    {
      "epoch": 0.15,
      "grad_norm": 6.0625,
      "learning_rate": 1.1178947368421054e-05,
      "loss": 0.2691,
      "step": 2920
    },
    {
      "epoch": 0.15,
      "grad_norm": 3.34375,
      "learning_rate": 1.1163157894736842e-05,
      "loss": 0.2865,
      "step": 2930
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.40625,
      "learning_rate": 1.1147368421052633e-05,
      "loss": 0.2395,
      "step": 2940
    },
    {
      "epoch": 0.15,
      "grad_norm": 3.921875,
      "learning_rate": 1.1131578947368421e-05,
      "loss": 0.2228,
      "step": 2950
    },
    {
      "epoch": 0.15,
      "grad_norm": 3.375,
      "learning_rate": 1.1115789473684212e-05,
      "loss": 0.2599,
      "step": 2960
    },
    {
      "epoch": 0.15,
      "grad_norm": 3.421875,
      "learning_rate": 1.11e-05,
      "loss": 0.2064,
      "step": 2970
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.34375,
      "learning_rate": 1.108421052631579e-05,
      "loss": 0.2317,
      "step": 2980
    },
    {
      "epoch": 0.15,
      "grad_norm": 3.4375,
      "learning_rate": 1.106842105263158e-05,
      "loss": 0.2188,
      "step": 2990
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.15625,
      "learning_rate": 1.1052631578947368e-05,
      "loss": 0.2675,
      "step": 3000
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.71875,
      "learning_rate": 1.1036842105263158e-05,
      "loss": 0.1853,
      "step": 3010
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.75,
      "learning_rate": 1.1021052631578947e-05,
      "loss": 0.2659,
      "step": 3020
    },
    {
      "epoch": 0.15,
      "grad_norm": 6.21875,
      "learning_rate": 1.1005263157894737e-05,
      "loss": 0.296,
      "step": 3030
    },
    {
      "epoch": 0.15,
      "grad_norm": 3.75,
      "learning_rate": 1.0989473684210526e-05,
      "loss": 0.2744,
      "step": 3040
    },
    {
      "epoch": 0.15,
      "grad_norm": 3.75,
      "learning_rate": 1.0973684210526316e-05,
      "loss": 0.2558,
      "step": 3050
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.53125,
      "learning_rate": 1.0957894736842105e-05,
      "loss": 0.3046,
      "step": 3060
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.46875,
      "learning_rate": 1.0942105263157894e-05,
      "loss": 0.2861,
      "step": 3070
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.1875,
      "learning_rate": 1.0926315789473684e-05,
      "loss": 0.237,
      "step": 3080
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.90625,
      "learning_rate": 1.0910526315789474e-05,
      "loss": 0.2734,
      "step": 3090
    },
    {
      "epoch": 0.15,
      "grad_norm": 3.84375,
      "learning_rate": 1.0894736842105265e-05,
      "loss": 0.3093,
      "step": 3100
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.25,
      "learning_rate": 1.0878947368421053e-05,
      "loss": 0.2508,
      "step": 3110
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.28125,
      "learning_rate": 1.0863157894736844e-05,
      "loss": 0.264,
      "step": 3120
    },
    {
      "epoch": 0.16,
      "grad_norm": 3.03125,
      "learning_rate": 1.0847368421052632e-05,
      "loss": 0.1805,
      "step": 3130
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.15625,
      "learning_rate": 1.0831578947368421e-05,
      "loss": 0.2598,
      "step": 3140
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.25,
      "learning_rate": 1.0815789473684211e-05,
      "loss": 0.2245,
      "step": 3150
    },
    {
      "epoch": 0.16,
      "grad_norm": 5.0625,
      "learning_rate": 1.08e-05,
      "loss": 0.2477,
      "step": 3160
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.8125,
      "learning_rate": 1.078421052631579e-05,
      "loss": 0.309,
      "step": 3170
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.34375,
      "learning_rate": 1.0768421052631579e-05,
      "loss": 0.2718,
      "step": 3180
    },
    {
      "epoch": 0.16,
      "grad_norm": 3.46875,
      "learning_rate": 1.075263157894737e-05,
      "loss": 0.2231,
      "step": 3190
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.375,
      "learning_rate": 1.0736842105263158e-05,
      "loss": 0.2859,
      "step": 3200
    },
    {
      "epoch": 0.16,
      "grad_norm": 5.9375,
      "learning_rate": 1.0721052631578947e-05,
      "loss": 0.276,
      "step": 3210
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.28125,
      "learning_rate": 1.0705263157894737e-05,
      "loss": 0.2616,
      "step": 3220
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.609375,
      "learning_rate": 1.0689473684210526e-05,
      "loss": 0.2865,
      "step": 3230
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.1875,
      "learning_rate": 1.0673684210526316e-05,
      "loss": 0.2276,
      "step": 3240
    },
    {
      "epoch": 0.16,
      "grad_norm": 3.8125,
      "learning_rate": 1.0657894736842106e-05,
      "loss": 0.2579,
      "step": 3250
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.5,
      "learning_rate": 1.0642105263157897e-05,
      "loss": 0.3116,
      "step": 3260
    },
    {
      "epoch": 0.16,
      "grad_norm": 3.546875,
      "learning_rate": 1.0626315789473685e-05,
      "loss": 0.2143,
      "step": 3270
    },
    {
      "epoch": 0.16,
      "grad_norm": 5.09375,
      "learning_rate": 1.0610526315789474e-05,
      "loss": 0.2593,
      "step": 3280
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.0625,
      "learning_rate": 1.0594736842105264e-05,
      "loss": 0.232,
      "step": 3290
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.71875,
      "learning_rate": 1.0578947368421053e-05,
      "loss": 0.229,
      "step": 3300
    },
    {
      "epoch": 0.17,
      "grad_norm": 3.921875,
      "learning_rate": 1.0563157894736843e-05,
      "loss": 0.2862,
      "step": 3310
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.140625,
      "learning_rate": 1.0547368421052632e-05,
      "loss": 0.3062,
      "step": 3320
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.625,
      "learning_rate": 1.0531578947368422e-05,
      "loss": 0.2486,
      "step": 3330
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.90625,
      "learning_rate": 1.051578947368421e-05,
      "loss": 0.3082,
      "step": 3340
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.15625,
      "learning_rate": 1.05e-05,
      "loss": 0.2006,
      "step": 3350
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.53125,
      "learning_rate": 1.048421052631579e-05,
      "loss": 0.246,
      "step": 3360
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.75,
      "learning_rate": 1.0468421052631578e-05,
      "loss": 0.201,
      "step": 3370
    },
    {
      "epoch": 0.17,
      "grad_norm": 3.84375,
      "learning_rate": 1.0452631578947369e-05,
      "loss": 0.1955,
      "step": 3380
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.375,
      "learning_rate": 1.0436842105263157e-05,
      "loss": 0.243,
      "step": 3390
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.53125,
      "learning_rate": 1.0421052631578948e-05,
      "loss": 0.2789,
      "step": 3400
    },
    {
      "epoch": 0.17,
      "grad_norm": 3.625,
      "learning_rate": 1.0405263157894736e-05,
      "loss": 0.1762,
      "step": 3410
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.84375,
      "learning_rate": 1.0389473684210527e-05,
      "loss": 0.214,
      "step": 3420
    },
    {
      "epoch": 0.17,
      "grad_norm": 3.78125,
      "learning_rate": 1.0373684210526317e-05,
      "loss": 0.2191,
      "step": 3430
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.8125,
      "learning_rate": 1.0357894736842106e-05,
      "loss": 0.292,
      "step": 3440
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.875,
      "learning_rate": 1.0342105263157896e-05,
      "loss": 0.2538,
      "step": 3450
    },
    {
      "epoch": 0.17,
      "grad_norm": 3.328125,
      "learning_rate": 1.0326315789473685e-05,
      "loss": 0.258,
      "step": 3460
    },
    {
      "epoch": 0.17,
      "grad_norm": 3.375,
      "learning_rate": 1.0310526315789475e-05,
      "loss": 0.2307,
      "step": 3470
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.828125,
      "learning_rate": 1.0294736842105264e-05,
      "loss": 0.2486,
      "step": 3480
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.8828125,
      "learning_rate": 1.0278947368421052e-05,
      "loss": 0.143,
      "step": 3490
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.0,
      "learning_rate": 1.0263157894736843e-05,
      "loss": 0.2726,
      "step": 3500
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.21875,
      "learning_rate": 1.0247368421052631e-05,
      "loss": 0.2666,
      "step": 3510
    },
    {
      "epoch": 0.18,
      "grad_norm": 3.59375,
      "learning_rate": 1.0231578947368422e-05,
      "loss": 0.2403,
      "step": 3520
    },
    {
      "epoch": 0.18,
      "grad_norm": 8.875,
      "learning_rate": 1.021578947368421e-05,
      "loss": 0.3262,
      "step": 3530
    },
    {
      "epoch": 0.18,
      "grad_norm": 6.9375,
      "learning_rate": 1.02e-05,
      "loss": 0.2611,
      "step": 3540
    },
    {
      "epoch": 0.18,
      "grad_norm": 6.84375,
      "learning_rate": 1.018421052631579e-05,
      "loss": 0.2457,
      "step": 3550
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.6875,
      "learning_rate": 1.0168421052631578e-05,
      "loss": 0.2983,
      "step": 3560
    },
    {
      "epoch": 0.18,
      "grad_norm": 3.984375,
      "learning_rate": 1.0152631578947368e-05,
      "loss": 0.2505,
      "step": 3570
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.234375,
      "learning_rate": 1.0136842105263157e-05,
      "loss": 0.2786,
      "step": 3580
    },
    {
      "epoch": 0.18,
      "grad_norm": 3.25,
      "learning_rate": 1.0121052631578949e-05,
      "loss": 0.3497,
      "step": 3590
    },
    {
      "epoch": 0.18,
      "grad_norm": 3.5,
      "learning_rate": 1.0105263157894738e-05,
      "loss": 0.2734,
      "step": 3600
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.25,
      "learning_rate": 1.0089473684210528e-05,
      "loss": 0.2685,
      "step": 3610
    },
    {
      "epoch": 0.18,
      "grad_norm": 8.3125,
      "learning_rate": 1.0073684210526317e-05,
      "loss": 0.3044,
      "step": 3620
    },
    {
      "epoch": 0.18,
      "grad_norm": 6.40625,
      "learning_rate": 1.0057894736842105e-05,
      "loss": 0.2584,
      "step": 3630
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.46875,
      "learning_rate": 1.0042105263157896e-05,
      "loss": 0.2911,
      "step": 3640
    },
    {
      "epoch": 0.18,
      "grad_norm": 3.859375,
      "learning_rate": 1.0026315789473684e-05,
      "loss": 0.2277,
      "step": 3650
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.3125,
      "learning_rate": 1.0010526315789474e-05,
      "loss": 0.2325,
      "step": 3660
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.21875,
      "learning_rate": 9.994736842105263e-06,
      "loss": 0.296,
      "step": 3670
    },
    {
      "epoch": 0.18,
      "grad_norm": 3.59375,
      "learning_rate": 9.978947368421053e-06,
      "loss": 0.2246,
      "step": 3680
    },
    {
      "epoch": 0.18,
      "grad_norm": 3.796875,
      "learning_rate": 9.963157894736842e-06,
      "loss": 0.2403,
      "step": 3690
    },
    {
      "epoch": 0.18,
      "grad_norm": 3.4375,
      "learning_rate": 9.94736842105263e-06,
      "loss": 0.2477,
      "step": 3700
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.6875,
      "learning_rate": 9.931578947368421e-06,
      "loss": 0.2639,
      "step": 3710
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.6875,
      "learning_rate": 9.91578947368421e-06,
      "loss": 0.2638,
      "step": 3720
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.5625,
      "learning_rate": 9.9e-06,
      "loss": 0.2519,
      "step": 3730
    },
    {
      "epoch": 0.19,
      "grad_norm": 3.34375,
      "learning_rate": 9.884210526315789e-06,
      "loss": 0.2468,
      "step": 3740
    },
    {
      "epoch": 0.19,
      "grad_norm": 3.453125,
      "learning_rate": 9.868421052631579e-06,
      "loss": 0.2896,
      "step": 3750
    },
    {
      "epoch": 0.19,
      "grad_norm": 3.796875,
      "learning_rate": 9.85263157894737e-06,
      "loss": 0.2912,
      "step": 3760
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.625,
      "learning_rate": 9.836842105263158e-06,
      "loss": 0.3592,
      "step": 3770
    },
    {
      "epoch": 0.19,
      "grad_norm": 3.375,
      "learning_rate": 9.821052631578948e-06,
      "loss": 0.289,
      "step": 3780
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.0625,
      "learning_rate": 9.805263157894737e-06,
      "loss": 0.168,
      "step": 3790
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.96875,
      "learning_rate": 9.789473684210527e-06,
      "loss": 0.2772,
      "step": 3800
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.625,
      "learning_rate": 9.773684210526316e-06,
      "loss": 0.2263,
      "step": 3810
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.5,
      "learning_rate": 9.757894736842106e-06,
      "loss": 0.3336,
      "step": 3820
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.0625,
      "learning_rate": 9.742105263157895e-06,
      "loss": 0.1924,
      "step": 3830
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.15625,
      "learning_rate": 9.726315789473684e-06,
      "loss": 0.2615,
      "step": 3840
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.28125,
      "learning_rate": 9.710526315789474e-06,
      "loss": 0.2536,
      "step": 3850
    },
    {
      "epoch": 0.19,
      "grad_norm": 3.890625,
      "learning_rate": 9.694736842105263e-06,
      "loss": 0.2672,
      "step": 3860
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.03125,
      "learning_rate": 9.678947368421053e-06,
      "loss": 0.2703,
      "step": 3870
    },
    {
      "epoch": 0.19,
      "grad_norm": 6.9375,
      "learning_rate": 9.663157894736842e-06,
      "loss": 0.2596,
      "step": 3880
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.96875,
      "learning_rate": 9.647368421052632e-06,
      "loss": 0.2443,
      "step": 3890
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.96875,
      "learning_rate": 9.63157894736842e-06,
      "loss": 0.2118,
      "step": 3900
    },
    {
      "epoch": 0.2,
      "grad_norm": 3.96875,
      "learning_rate": 9.61578947368421e-06,
      "loss": 0.2678,
      "step": 3910
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.59375,
      "learning_rate": 9.600000000000001e-06,
      "loss": 0.2817,
      "step": 3920
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.59375,
      "learning_rate": 9.58421052631579e-06,
      "loss": 0.225,
      "step": 3930
    },
    {
      "epoch": 0.2,
      "grad_norm": 6.1875,
      "learning_rate": 9.56842105263158e-06,
      "loss": 0.2452,
      "step": 3940
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.84375,
      "learning_rate": 9.552631578947369e-06,
      "loss": 0.1778,
      "step": 3950
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.96875,
      "learning_rate": 9.53684210526316e-06,
      "loss": 0.312,
      "step": 3960
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.5625,
      "learning_rate": 9.521052631578948e-06,
      "loss": 0.2525,
      "step": 3970
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.03125,
      "learning_rate": 9.505263157894737e-06,
      "loss": 0.2373,
      "step": 3980
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.8125,
      "learning_rate": 9.489473684210527e-06,
      "loss": 0.2765,
      "step": 3990
    },
    {
      "epoch": 0.2,
      "grad_norm": 3.171875,
      "learning_rate": 9.473684210526315e-06,
      "loss": 0.2565,
      "step": 4000
    },
    {
      "epoch": 0.2,
      "eval_loss": 0.2541459798812866,
      "eval_runtime": 64.6611,
      "eval_samples_per_second": 15.465,
      "eval_steps_per_second": 15.465,
      "step": 4000
    }
  ],
  "logging_steps": 10,
  "max_steps": 10000,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 2000,
  "total_flos": 6.455688167424e+16,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}